mirror of
https://github.com/openai/codex.git
synced 2026-11-17 22:10:37 +00:00
## What changed - Open PNG, JPEG, GIF, and WebP file links as sidebar image previews, fitted to the available space with format and dimension labels. Show a fallback when terminal graphics are unavailable, and preserve text previews for files such as Git LFS pointers. - In the owned transcript, place completed, saved image results after the assistant's answer, with larger, left-aligned previews and a clickable `Generated image` label. Preserve this ordering during replay and history pagination, including interrupted turns. - Reuse native Kitty image IDs and move existing placements without retransmitting their pixels. Probe Kitty graphics support in `herdr` before enabling previews. ## Testing Add regression coverage for sidebar preview fallbacks, generated-image ordering, pagination and search, preview layout, Kitty placement reuse, and graphics probe replies interleaved with pasted input. GitOrigin-RevId: 22d8aa2724f17bfa7f8280a2d11d4a7e2cb66728